Detecting passing valves

Abstract

This disclosure describes systems and methods for detecting passing valves. A method includes acquiring vibrational data from one or more sensors associated with passing valves and non-passing valves; extracting a plurality of features from the vibrational data; determining, based on a feature importance criterion, a subset of the plurality of features having more significance than other features of the plurality of features; training a machine learning model, where inputs to the machine learning model include the set of features; detecting that a valve is a passing valve based on the trained machine learning model, where an input to the trained machine learning model includes the subset of features extracted from vibrational data; and in response to detecting the passing valve, performing a corrective action to resolve the passing valve.

What is claimed is: 
     
         1 . A method for detecting passing valves, the method comprising:
 acquiring vibrational data from one or more sensors associated with passing valves and non-passing valves;   extracting a plurality of features from the vibrational data;   determining, based on a feature importance criterion, a subset of the plurality of features having more significance than other features of the plurality of features;   training a machine learning model, where inputs to the machine learning model include the set of features;   detecting that a valve is a passing valve based on the trained machine learning model, where an input to the trained machine learning model includes the subset of features extracted from vibrational data; and   in response to detecting the passing valve, performing a corrective action to resolve the passing valve.   
     
     
         2 . The method of  claim 1 , wherein the corrective action comprises at least one of generating an alert indicating the detection of the passing valve or automatically closing a valve upstream of the detected passing valve. 
     
     
         3 . The method of  claim 1 , wherein the one or more sensors comprise one or more analog piezoelectric vibrational sensors. 
     
     
         4 . The method of  claim 3 , wherein extracting a plurality of features comprises:
 filtering the vibrational data using a bandpass filter; and   converting the filtered vibrational data to digital vibrational data using a high-sampling rate analog to digital converter.   
     
     
         5 . The method of  claim 4 , wherein the sampling rate of the analog to digital converter is at least 2 MHz. 
     
     
         6 . The method of  claim 4 , wherein the bandpass filter passes frequencies between 100 kHz and 300 kHz. 
     
     
         7 . The method of  claim 1 , wherein extracting the plurality of features from the vibrational data includes determining one or more of a root mean square value, a spectral roll off, a spectral bandwidth, a zero-crossing rate, and Mel-Frequency Cepstral Coefficients. 
     
     
         8 . The method of  claim 7 , wherein the feature importance criterion comprises a reduction in a percentage of results classified correctly when a feature is omitted; and
 a feature having more significance has a higher reduction in the percentage of results classified correctly when the feature is omitted relative to the reduction in the percentage of results when other features are omitted.   
     
     
         9 . The method of  claim 1 , wherein acquiring vibrational data associated with passing valves and non-passing valves comprises:
 acquiring, from a testing device, the vibrational data associated with multiple valve types and multiple pipe diameters.   
     
     
         10 . A system for detecting passing valves, the system comprising:
 one or more piezoelectric sensors coupled to a pipe adjacent to a valve;   at least one processor and a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
 acquiring vibrational data from the one or more piezo electric sensors associated with passing valves and non-passing valves; 
 extracting a plurality of features from the vibrational data; 
 determining, based on a feature importance criterion, a subset of the plurality of features having more significance than other features of the plurality of features; and 
 training a machine learning model, where inputs to the machine learning model include the set of features. 
   
     
     
         11 . The system of  claim 10 , wherein the operations further comprise:
 detecting that a valve is a passing valve based on the trained machine learning model, where an input to the trained machine learning model includes the subset of features extracted from vibrational data; and   in response to detecting the passing valve, performing a corrective action to resolve the passing valve.   
     
     
         12 . The system of  claim 11 , wherein the corrective action comprises at least one of generating an alert indicating the detection of the passing valve or causing a valve upstream of the detected passing valve to close automatically. 
     
     
         13 . The system of  claim 10 , wherein extracting the plurality of features from the vibrational data includes determining one or more of a root mean square value, a spectral roll off, a spectral bandwidth, a zero-crossing rate, and Mel-Frequency Cepstral Coefficients. 
     
     
         14 . The system of  claim 13 , wherein the feature importance criterion comprises a reduction in a percentage of results classified correctly when a feature is omitted; and
 a feature having more significance has a higher reduction in the percentage of results classified correctly when the feature is omitted relative to the reduction in the percentage of results when other features are omitted.   
     
     
         15 . The system of  claim 10 , wherein the one or more piezoelectric sensors comprise one or more analog piezoelectric vibrational sensors, and
 wherein the operations further comprise:
 filtering the vibrational data using a bandpass filter; and 
 converting the filtered vibrational data to digital vibrational data using a high-sampling rate analog to digital converter. 
   
     
     
         16 . One or more non-transitory machine-readable storage devices storing instructions for detecting passing valves, the instructions being executable by one or more processors, to cause performance of operations comprising:
 acquiring vibrational data from the one or more piezo electric sensors associated with passing valves and non-passing valves;   extracting a plurality of features from the vibrational data;   determining, based on a feature importance criterion, a subset of the plurality of features having more significance than other features of the plurality of features; and   training a machine learning model, where inputs to the machine learning model include the set of features.   
     
     
         17 . The non-transitory machine-readable storage devices of  claim 16 , wherein the operations further comprise:
 detecting that a valve is a passing valve based on the trained machine learning model, where an input to the trained machine learning model includes the subset of features extracted from vibrational data; and   in response to detecting the passing valve, performing a corrective action to resolve the passing valve.   
     
     
         18 . The non-transitory machine-readable storage devices of  claim 17 , wherein the corrective action comprises at least one of generating an alert indicating the detection of the passing valve or causing a valve upstream of the detected passing valve to close automatically. 
     
     
         19 . The non-transitory machine-readable storage devices of  claim 16 , wherein extracting the plurality of features from the vibrational data includes determining one or more of a root mean square value, a spectral roll off, a spectral bandwidth, a zero-crossing rate, and Mel-Frequency Cepstral Coefficients. 
     
     
         20 . The non-transitory machine-readable storage devices of  claim 19 , wherein the feature importance criterion comprises a reduction in a percentage of results classified correctly when a feature is omitted; and
 a feature having more significance has a higher reduction in the percentage of results classified correctly when the feature is omitted relative to the reduction in the percentage of results when other features are omitted.